Autonomous Intersections
(Deep Q Learning)

Personal Project

In progress

This is a small project that stemmed from a hackathon idea. The big picture idea is to show how autonomous cars could eliminate the need for traffic lights, even if not all cars are autonomous. Thus the goal is to develop a collision avoidance algorithm to control the autonomous cars, even if there is a user controlled car that is malicious (trying to crash into other cars). 

​

In this simulation, there are only one-lane roads (say 3 vertical 3 horizontal) that simulate a city grid.

​

I found that Deep Q Learning might be an elegant solution to the complicated problem. So I am looking forward to doing more research and implementing a DQN to solve the problem.